Prob with powerpath - powermt config

[Stephanie Nicholls]

Hi

we have a rx4640 running v11.23 connected to a Clariion CX700 via a fabric. The machine has 2 HBA's and powerpath installed. Currently powerpath is recognising only one HBA and directing all I/O down those paths. At a HPUX level there appears to be nothing wrong. An ioscan shows LUNs down both HBA's. fcmsutil returns seem normal and show both HBA's logged into the fabric, and if I do a vgextend I can add the alternate path to a volume group, and the I/O count on the fcmsutil stats goes up. But 'powermt config' refuses to pick up the other HBA. Any ideas on how to force it ? I would prefer not to have to reboot the machine if I can avoid it.

powermt display unmanged shows no unmanaged devices either.

Any ideas ??

Thanks

Steph

PS: Oh yeah, it used to work and stopped - not sure when as I only just got back from long term leave.

[D Block 2]

11.23 hpux -- what version of Powerpath have you installed ? I would start here.

also, are the Sym SER#'s the same for the primary and alternate disk devices using: INQ

I don't think you need to do any powermt commands once you install powerpath.. it should see the vgextended or Alternative links and manager accordingly.

also, wonder.. does the ECC control center see both devices ? if not maybe a Zoning issue ?

[Mark Coleman_1]

Hi Stephanie,
Do a powermt display dev=all | more and look through that to make sure all the paths are alive. If so, also take a look at the default owner of the LUN (SP A or B) and the current owner. Sometimes the LUNS will trespass and only show up on one device.
If you have some dead paths show up from the powermt display command, do a powermt check force to get rid of the dead paths, then an ioscan, then insf -e -C disk and then powermt config again. That should get all your paths back active and powerpath splitting the IO out correctly.
You can do the ioscan and insf commands anyway if you feel inclined. Normally powerpath won't show the LUN unless it has all it's device files.
As a last resort you can stop and restart the Navisphere agent with the /sbin/init.d/agent stop/start command. I've never really had much luck with that, but I've been told to do it a few times.

Good luck!
Mark

[Raj D.]

Hi Steph ,

You can run

# powermt check
# powermt config
# powermt check
# powermt display
# powermt save

You can check and correct if there is any dead path.

Cheers,
Raj.

[Raj D.]

Hi Steph ,

You can run

# powermt check
# powermt config
# powermt check
# powermt display
# powermt save

It will= check and correct if there is any dead path.

Cheers,
Raj.

[Stephanie Nicholls]

Hi folks, thanks for the responses so far. The problem is NOT a dead path, the problem is powermt display only shows ONE HBA not TWO HBA's

a powermt display shows,
root@xxxxxxx:#>powermt display
Symmetrix logical device count=0
CLARiiON logical device count=19
==============================================================================
----- Host Bus Adapters --------- ------ I/O Paths ----- ------ Stats ------
### HW Path Summary Total Dead IO/Sec Q-IOs Errors
==============================================================================
10 0/3/2/0.1.16.0.0 optimal 19 0 - 0 0
11 0/3/2/0.1.31.0.0 optimal 19 0 - 0 57


However the IOSCAN shows LUNS down the hardware path
0/3/1/0.2.16.0 & 0/3/1/0.2.31.0 as well.

but I cant work out how to get powerpath to recognise those devices. (They are alternate links, but I need powerpath to see them for the i/o load balancing and redundancy.)

Cheers

Steph

[Mark Coleman_1]

OK. Might be a zoning issue. Not sure what types of fiber switches you have, but check that the second HBA is zoned to SP A and B of the Clariion.

Mark

[Bernd Reize · Accepted solution]

Zoning should be correct, since ioscan shows all paths corretly. Its only a powerpath problem, if I understand right.

Can you please check the initiator settings on the Clariion? To have powerpath recognize the devices the initiator type should be set to "HP No Auto Trespass" Maybe the second HBA ist set to the wrong type?

[Stephanie Nicholls]

Thanks Bernd, for your suggestion it was absolutely spot on. I suspect someone changed this value when they did the flare upgrade last month.

Many thanks

Steph

PS: This saved me from one of those support calls you never ever want to make, what bright spark in the company bought powerpath for HPUX from Dell ?!?

[Karim Abdullah]

Hi, we are having some similar issues.

We have 1 sun and 1 HP server using Powerpath / connected to SAN / Clariion CX700.
We need to upgrade the Clariion to flare code 19.
EMC is saying the power path versions has to be upgraded first.
Can we shut down the servers, perform the flare code upgrade and then bring the servers back up...anyone with similar scenario ?